Digital literacy and scientific research ability among nursing postgraduates: The chain-mediating role of deep learning and research self-efficacy.

BACKGROUND: In the era of rapid digital transformation, it is crucial to cultivate research competence among nursing postgraduates to support innovation in nursing science and strengthen the overall competitiveness of the profession. However, limited attention has been given to how digital literacy is associated with perceived (self-reported) research competence and whether this association can be statistically accounted for by deep learning and research self-efficacy. OBJECTIVES: This study examined the association between digital literacy and perceived research competence among nursing postgraduates and estimated indirect effect (statistical) pathways through deep learning and research self-efficacy in a hypothesized serial (chain) mediation model. DESIGN: A cross-sectional study design was utilized. SETTINGS: This multi-center study was carried out at eight universities located in three Chinese provinces: Anhui, Jiangsu, and Sichuan. PARTICIPANTS: Eligible nursing postgraduates from the aforementioned universities were recruited during the period of June to July 2025. METHODS: Data were collected using the Digital Literacy Scale, the Deep Learning Scale, the Research Self-Efficacy Scale, and the Research Competence Scale (self-reported perceived competence). Descriptive statistics, Pearson's correlation analysis, and serial mediation analyses were conducted using SPSS version 26.0 and the PROCESS macro (version 3.5). RESULTS: The mean score for perceived research competence among nursing postgraduates was 39.88 ± 8.10 points. The mean scores for digital literacy, deep learning, and research self-efficacy were 85.37 ± 11.44, 40.54 ± 6.64, and 18.34 ± 3.46 points, respectively. Perceived research competence was significantly and positively correlated with digital literacy (r = 0.598, p < 0.001), deep learning (r = 0.549, p < 0.001), and research self-efficacy (r = 0.523, p < 0.001). Serial mediation analysis (PROCESS Model 6) yielded statistically significant indirect effect estimates through deep learning (21.71%), through research self-efficacy (15.33%), and through the serial pathway involving both mediators (7.60%). Across pathways, the total indirect effect estimate accounted for 44.58% of the total association estimate, whereas the direct effect estimate accounted for 55.42%, suggesting that digital literacy is linked to perceived research competence largely through direct research-related digital behaviors while a substantial proportion is statistically accounted for by deep learning and research self-efficacy. CONCLUSIONS: The findings are consistent with a model in which higher digital literacy is associated with higher levels of deep learning and research self-efficacy, which in turn are associated with higher perceived research competence. Given the cross-sectional, self-report design, the serial mediation model should be interpreted as a statistical decomposition of contemporaneous associations rather than evidence of temporal ordering or causal relationships. Importantly, the outcome reflects perceived (self-reported) competence and was not corroborated by objective performance indicators (e.g., publications, project outputs, or externally assessed research skills). Nursing education institutions may consider integrating digital literacy training, deep learning-oriented pedagogy, and supports for research self-efficacy into postgraduate curricula, while evaluating their causal impacts and objective outcomes in longitudinal or intervention studies. The generalisability of these findings may be limited because participants were recruited from eight universities in three provinces; replication in more diverse regions and institution types (including less-resourced settings) is warranted.